Crystal design approaches for the synthesis of paracetamol co-crystals

Article

Accepted Version

Srirambhatla, V. K., Kraft, A., Watt, S. and Powell, A. V. (2012) Crystal design approaches for the synthesis of paracetamol co-crystals. Crystal Growth & Design, 12 (10). pp. 4870-4879. ISSN 1528-7483 doi: https://doi.org/10.1021/cg300689m Available at http://centaur.reading.ac.uk/34128/

It is advisable to refer to the publisher’s version if you intend to cite from the work. See Guidance on citing .

To link to this article DOI: http://dx.doi.org/10.1021/cg300689m

Publisher: American Chemical Society

All outputs in CentAUR are protected by Intellectual Property Rights law, including copyright law. Copyright and IPR is retained by the creators or other copyright holders. Terms and conditions for use of this material are defined in the End User Agreement .

www.reading.ac.uk/centaur

CentAUR

Central Archive at the University of Reading

Reading’s research outputs online

Crystal Design Approaches for the Synthesis of Paracetamol Co-CrystalsVijay K. Srirambhatla,† Arno Kraft,† Stephen Watt,‡ and Anthony V. Powell*,†

†Department of Chemistry, School of Engineering and Physical Sciences, Heriot-Watt University, Edinburgh EH14 4AS, U.K.‡Solid Form Solutions Ltd, The Fleming Building, Edinburgh Technopole, Milton Bridge, Penicuik, EH26 0BE, U.K.

*S Supporting Information

ABSTRACT: Crystal engineering principles were used to designthree new co-crystals of paracetamol. A variety of potential co-crystal formers were initially identified from a search of theCambridge Structural Database for molecules with complemen-tary hydrogen-bond forming functionalities. Subsequent screen-ing by powder X-ray diffraction of the products of the reaction ofthis library of molecules with paracetamol led to the discovery ofnew binary crystalline phases of paracetamol with trans-1,4-diaminocyclohexane (1); trans-1,4-di(4-pyridyl)ethylene (2); and 1,2-bis(4-pyridyl)ethane (3). The co-crystals werecharacterized by IR spectroscopy, differential scanning calorimetry, and 1H NMR spectroscopy. Single crystal X-ray structureanalysis reveals that in all three co-crystals the co-crystal formers (CCF) are hydrogen bonded to the paracetamol moleculesthrough O−H···N interactions. In co-crystals (1) and (2) the CCFs are interleaved between the chains of paracetamol molecules,while in co-crystal (3) there is an additional N−H···N hydrogen bond between the two components. A hierarchy of hydrogenbond formation is observed in which the best donor in the system, the phenolic O−H group of paracetamol, is preferentiallyhydrogen bonded to the best acceptor, the basic nitrogen atom of the co-crystal former. The geometric aspects of the hydrogenbonds in co-crystals 1−3 are discussed in terms of their electrostatic and charge-transfer components.

■ INTRODUCTION

Co-crystals, or multicomponent molecular crystals,1 includeaddition compounds,2 organic molecular complexes,3 and solid-state complexes.4 The definition of a co-crystal is a subject ofdebate.1,5,6 While there is no consensus over the precisedefinition of what constitutes a co-crystal, a broad generality isthat that they are crystalline materials comprising at least twodifferent molecular components. Pharmaceutical co-crystalsrepresent a subset of this broad class, in which at least one ofthe components is an active pharmaceutical ingredient (API). Itis well established that different forms of a given pharmaceuti-cally active compound may have different physical properties,7

salt formation being one of the common approaches tomodifying the physical properties of a drug. However, saltformation is feasible only when the API possesses a suitableionizable site. In contrast, co-crystal formation exploits the veryfeatures that lead to a compound being an API: the presence offunctional groups. Moreover the functional groups may allowAPIs to exhibit polymorphism,8 exist as solvates,9 or form co-crystals.10 Given the importance of APIs, it is perhaps surprisingthat only in recent years have crystal engineering concepts beenapplied to APIs11−15 to enhance the range of potentially usableforms of a given drug substance.Crystal engineering can be defined as the rational design of

functional molecular solids.16 The concept was introduced byPepinsky17 and applied by Schmidt18 to organic solid-statephotochemical reactions. Thomas et al.19 in reviewing thereactivity of this category of reaction noted that the reactivity in

the solid state is governed by the precise control of molecularorientation and stacking sequences in the crystal structures.This observation paved the way for the further design andsynthesis of new solids based on crystal engineering strategies.The basic understanding of crystal engineering concepts isbased on analysis of the manner in which molecules are packedin molecular crystals also considered as supermolecules.20 Thecrystal structure of any organic compound may be consideredto be the result of a series of competing, directional, andpredictable interactions between functional groups, giving riseto recognizable structural building units termed supramolecularsynthons.21 Identifying reliable and robust supramolecularsynthons is key for effective crystal engineering. Shattock etal.22 have demonstrated that database mining using theCambridge Structural Database (CSD) provides valuableinformation on the occurrence of a particular supramolecularsynthon in series of closely related structures, which may beapplied to the synthesis of a wide range of co-crystals. The totalnumber of reported co-crystal structures in the CSD remainsrelatively low (8552 entries, ca. 3.3% of all organic entries) incomparison to the total number of organic structures (254475entries).23 Database mining has been shown to be an effectiveapproach to the search for hydrogen-bonded synthons for theconstruction of organic crystal structures.24

Received: May 20, 2012Revised: August 30, 2012

Article

pubs.acs.org/crystal

© XXXX American Chemical Society A dx.doi.org/10.1021/cg300689m | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Paracetamol or acetaminophenol is a widely used analgesicand antipyretic drug that is commonly used for the relief offever, headaches, and other minor aches and pains. It belongs toa class of drugs known as aniline analgesics. In its crystallinestate paracetamol commonly exists in two polymorphicforms.25 The monoclinic form I (Figure 1a,b) is considered

to be the thermodynamically stable form. Form II whichcrystallizes in the orthorhombic system (Figure 1c) is ofinterest in formulations because of its superior compressionproperties compared to that of form I. Although a thirdpolymorph (form III) has been identified,26 until recently itscrystal structure27 remained elusive. The paracetamol moleculeconsists of a benzene ring substituted with hydroxyl andacetamide groups in the para (1,4) positions. Both of thestructurally characterized forms of paracetamol exhibit a similarpattern of hydrogen bonds involving the phenolic O−H group,which acts an acceptor in N−H···O interactions and as a donorin O−H···OC interactions, to generate in both forms I andII, a two-dimensional network. Using the graph set approach tohydrogen bonding of Bernstein et al.28 both forms I and II maybe described as having C(7) and C(9) motifs. However, thetwo polymorphs of paracetamol differ in the orientation of themolecules. In form I paracetamol molecules are arranged in a

herringbone pattern to form puckered hydrogen-bonded sheets,whereas in form II, they form flat hydrogen-bonded sheets.In addition to the reported polymorphic forms, a number of

research groups have synthesized and characterized several co-crystal/solvate forms of paracetamol. Karki et al.29 synthesizedfour co-crystals of paracetamol involving theophylline, oxalicacid, naphthalene, and phenazine as CCFs. They alsodemonstrated that the co-crystal forms possess bettercompression properties compared to that of form II. Inaddition to these pharmaceutically acceptable co-crystals, multi-component crystalline forms (both two-component co-crystalsand solvates) of paracetamol containing N,N-dimethylpiper-azine, 4,4′-bipyridine, N-methylmorpholine, morpholine,1,4,8,11-tetraazacyclotetradecane, 1,4-diazabicyclo[2.2.2]-octane, ethanol, water, piperazine, and 1,4-dioxane as CCFsor solvate have been reported.30 In all the reported multi-component crystals of paracetamol the hydrogen bondsinvolving the donor groups, i.e., the phenolic O−H and theamidic N−H participate in conventional hydrogen bondsthrough O−H···N/O and N−H···O/N interactions involvingacceptors from either the paracetamol molecules or CCFs.Similarly the acceptor groups are involved in O···H−O/N typeintermolecular interactions.If we consider form I, the strongest hydrogen bonds are

formed between the phenolic O−H donor and the amideoxygen acceptor (d(H···O) = 1.77 Å), while those between theN−H donor and O−H acceptor are weaker (d(H···O) = 2.00Å). Therefore, in order to make co-crystals of paracetamol, it isnecessary to break these hydrogen bonds and introduceinteractions with the CCF that are of comparable strength. Inseeking to identify suitable co-crystal formers (CCFs) topromote such interactions, we performed a search of theCambridge Structural Database (CSD) for molecules contain-ing functional groups that form hydrogen bonds with similargeometrical parameters. The results of this targeted searchsuggested a series of molecules in which these functional groupsare present. These molecules were then screened for co-crystalformation. This led ultimately to the synthesis of the three newco-crystals of paracetamol with trans-1,4-diaminocyclohexane,1,2-bis(4-pyridyl)ethane, and trans-1,4-di(4-pyridyl)ethylene(Figure 2) reported here.

■ EXPERIMENTAL SECTIONCSD Analysis. Identification of candidate co-crystal formers

suitable for paracetamol was achieved by carrying out an analysis ofstructural information contained in the Cambridge Structural Database

Figure 1. The packing arrangement of form I of paracetamol viewedalong (a) [010] and (b) [100] and (c) form II viewed along [001].Solid red lines represent intermolecular hydrogen bonds. Key:nitrogen, blue circles; oxygen, red circles; carbon, black circles;hydrogen, small white circles.

Figure 2. Chemical structures of the active pharmaceutical ingredientand cocrystal formers investigated in this work.

Crystal Growth & Design Article

dx.doi.org/10.1021/cg300689m | Cryst. Growth Des. XXXX, XXX, XXX−XXXB

Table 1. The Potential Co-Crystal Formers Used for Screening with Paracetamol

Crystal Growth & Design Article

dx.doi.org/10.1021/cg300689m | Cryst. Growth Des. XXXX, XXX, XXX−XXXC

(CSD). The search was performed using CSD version 5.33(November 2011) + 2 updates. The following search criteria wereapplied: only those organic structures for which three-dimensionalstructural coordinates have been determined were considered; onlythose crystal structures for which R ≤ 7.5% were retained in the resultsof the search; structures which showed evidence of disorder or hadbeen shown to contain errors were excluded, as were those forpolymeric species.All starting materials were obtained from Sigma-Aldrich and were

used as obtained. The solvent-drop grinding method31 was used toscreen for co-crystal formation. Stoichiometric amounts of para-cetamol and potential CCF were ground together with a mortar andpestle for ca. 5 min following addition of solvent (15 μL of solvent per50 mg of starting materials). The grinding experiments wereperformed in seven different solvents, methanol, ethanol, acetonitrile,chloroform, acetone, tetrahydrofuran, and n-propanol, representing awide range of dielectric constants (4.8−37.5) and containing at leastone representative from each of the polar protic, polar aprotic, andnonpolar solvent categories. The resulting solid material was analyzedby powder X-ray diffraction (PXRD). A total of 168 liquid assistedgrinding experiments were conducted. The CCFs and solvent systemsinvestigated are presented in Table 1.Analysis of the X-ray diffraction data of the polycrystalline materials

arising from these solvent drop-grinding experiments revealed that forcritical paracetamol/CCF ratios, reflections arising from the startingmaterials are absent, indicating the presence of new phases, whentrans-1,4-diaminocyclohexane, 1,2-bis(4-pyridyl)ethane, and 1,2-di(4-pyridyl)ethylene are used as CCFs. These molecules weresubsequently used to prepare single-crystal samples by solventevaporation methods, using the ratios for which reflections arisingfrom the starting materials were absent from PXRD data.Crystallization experiments were conducted in various combinationsof commonly occurring organic solvents; however, only thoseexperiments which led to diffraction quality crystals are detailedbelow. Single crystals of paracetamol/trans-1,4-diaminocyclohexane(1) were prepared from a solution of paracetamol (302 mg, 2.00mmol) and trans-1,4-diaminocyclohexane (114 mg, 1.00 mmol) in 3mL of methanol. The solution was heated for 3 min before beingallowed to stand at room temperature. After 4 days colorless crystals of(1) appeared. Pale yellow colored crystals of paracetamol:1,2-di(4-pyridyl)ethylene (2) were obtained by evaporation, over 3 days, ofsolvent from a solution containing 2 mmol (302 mg) of paracetamoland 1 mmol of 1,2-di(4-pyridyl)ethylene (182 mg) in 5 mL ofmethanol. A similar method was used for the preparation ofparacetamol/1,2-bis(4-pyridyl)ethane (3) from a 1:1 molar ratio ofparacetamol (151 mg, 1 mmol)/1,2-bis(4-pyridyl)ethane (184 mg,1.00 mmol) in 2 mL of dimethyl sulphoxide. Crystals emerged afterallowing the solution to stand at room temperature for 5 days.Polycrystalline materials obtained from the crystallization experi-

ments were analyzed by PXRD and compared with powder diffractionpatterns generated from the structures determined by single-crystal

methods. Diffraction data for the powders resulting from the use oftrans-1,4-diaminocyclohexane and 1,2-bis(4-pyridyl)ethane as CCFsare consistent with a bulk phase of co-crystals (1) and (3) respectively,irrespective of the solvent used. In the case where trans-1,4-di(4-pyridyl)ethylene was used as a CCF the PXRD pattern matches thesimulated powder diffraction pattern generated from the single crystalstructure of (2) when the solvent of crystallization is methanol.However, the powder diffraction data for products of crystallizationexperiments in other solvents do not correspond to the simulatedpattern of (2) and show a dependence on the solvent used. This mayindicate formation of alternative solvated co-crystal phases, althoughdiffraction quality crystals could not be obtained.

PXRD data for the polycrystalline products of screening for CCFformation were collected using a Bruker D8 Discover diffractometer,operating in transmission geometry with Cu Kα radiation (λ = 1.5418Å) and fitted with a Gobel X-ray focusing mirror and a LynxEye lineardetector. Finely ground samples of each of the products of screeningwere contained in a 96-well plate. Data for each sample were collectedover the angular range 4 ≤ 2θ/° ≤ 50, counting for 0.25 s at each0.02019 increment of the detector position.

Single crystal X-ray diffraction data for the co-crystals were collectedusing a Bruker Nonius X8-Apex2 CCD diffractometer operating withgraphite-monochromatized Mo Kα radiation (λ = 0.710173 Å). AnOxford Cryosystems Cryostream was used to cool the crystals to 100K prior to data collection. Data were processed using themanufacturer’s standard routines.32 The crystal structures were solvedby direct methods using the program SHELXS,33 and subsequentFourier calculations and least-squares refinements were performed onF using the program CRYSTALS.34 All non-hydrogen atoms wererefined with anisotropic displacement parameters. All hydrogen atomsbonded to the carbon atoms were placed geometrically and refinedwith the isotropic displacement parameter fixed at 1.5 times Ueq of theatoms to which they are attached. Protons involved in hydrogenbonding were located directly via inspection of difference Fourier mapsand were refined isotropically.

Infrared spectroscopic data for co-crystals 1−3 were obtained usinga Perkin-Elmer 100 ATR spectrometer. Ground crystals of co-crystals1−3 were placed on the crystal plate of the infrared instrument andthe spectrum was recorded in diffuse reflectance geometry. Differentialscanning calorimetric data for the samples 1−3 were obtained using aSeiko model DSC-6200 instrument. The samples were weighed intoaluminum DSC pans and crimped with a pinhole on the lid. Thesamples were analyzed over the temperature range 25−280 °C, using aheating rate of 10 °C. The data were processed using MuseMeasurement v. 5.4 (Build 263) software.

Samples for 1H NMR were dissolved in DMSO-d6 and spectraobtained using a Bruker 300 MHz spectrometer. The 1H NMRspectrum was recorded to confirm the API/CCF ratio in the bulkmaterial. 1H chemical shift values are reported on the δ scale . 1HNMR (300 MHz, DMSO-d6) of the co-crystal (1): δ 9.64 (s, 2H), 7.32and 6.66 (AA′XX′, 2 × 4H), 3.97 (br. s, 6H), 2.45 (m, 2H), 1.97 (s,

Table 1. continued

Crystal Growth & Design Article

dx.doi.org/10.1021/cg300689m | Cryst. Growth Des. XXXX, XXX, XXX−XXXD

6H), 1.68 (m, 4H), 1.00 (m, 4H). 1H NMR (DMSO-d6) of co-crystal(2): 9.65 (s, 2H), 9.16 (br. s, 2H), 8.60 and 7.60 (AA′XX′, 2 × 4H),7.53 (s, 2H), 7.34 and 6.68 (AA′XX′, 2 × 4H), 4.13 (s, 1H), 3.17 (s,3H), 1.97 (s, 6H). 1H NMR (DMSO-d6) of co-crystal (3): δ 9.66 (s,2H), 9.15 (s, 2H), 8.44 and 7.25 (AA′XX′, 2 × 4H), 7.33 and 7.25(AA′XX′, 2 × 4H), 2.94 (s, 8H), 1.97 (s, 6H).

■ RESULTS AND DISCUSSION

A summary of the results of the search of the CSD is presentedin Table 2. The database search was conducted to findcomplementary co-crystal formers targeting the phenolic OH−group of paracetamol. The results of the CSD search arepresented in two categories: raw and refined. The former refersto the search results for a specified functional group(s)including those occurrences where other functional groupsmay be present, whereas the latter contains those results forwhich there are no additional functional group(s) in theimmediate vicinity of those for which the search was conducted.The CSD search reveals 8009 crystal structures containing atleast one phenolic O−H moiety of which 13% exhibit theOH···OH supramolecular homosynthon. However, the remain-der of the crystal structures in the search results are dominatedby supramolecular heterosynthons. The CSD analysis revealsthat of those structures containing a phenolic hydroxyl grouptogether with functional groups such as aromatic nitrogenatoms and amines, 42−52% involve supramolecular hetero-synthon formation, even in the presence of other functionalgroups. In the absence of other hydrogen bond donors oracceptors the percentages rise to 54% and 64% for aromaticnitrogen atoms and amines respectively. The search also revealsthat for interactions of phenols with amides, in those caseswhere homosynthons are observed, the CONH2···CONH2homosynthon dominates over the OH···OH homosynthon, inboth the raw and refined data. However, heterosynthonsinvolving O−H/amide interactions are predominant in therefined data. For the interactions of phenols with carboxylicacids, OH···OC(OH) supramolecular heterosynthons exhibit56% reliability in the presence of competing functional groups,which increases to 67% in the absence of competing functionalgroups, while the analysis of phenol−amine interactionsindicates that the OH···NH2 supramolecular heterosynthon ismore common than the OH···OH supramolecular homosyn-thon. In summary, analysis of the CSD for structures in whichthere are noncovalent interactions between phenolic and othercomplementary functional groups reveals a marked preferencefor supramolecular heterosynthons over supramolecular homo-synthons. Taking into account the percentage occurrence ofsupramolecular heterosynthons observed in the refinedsearches, molecules containing at least one of the functionalgroups of aromatic nitrogen group, amides, carbonyl groups,carboxylic acids, and amines were selected as potential co-crystal formers for screening studies (Table 1).PXRD of the solids obtained from the liquid assisted

grinding method of screening revealed physical mixtures ofstarting materials in most cases. In such cases the liquid phasemay merely serve as a lubricant, which would be consistent withthe lack of any dependence of the nature of the solid producton the choice of solvent. However, the reflections arising fromthe starting materials were absent when trans-1,4-diaminocy-clohexane, 1,2-bis(4-pyridyl)ethane, and 1,2-di(4-pyridyl)-ethylene are used as CCFs. PXRD data for the products ofsolvent-drop grinding experiments using 1,2-di(4-pyridyl)-ethylene as CCF reveal that while reflections of the starting Table

2.Summaryof

theResultsfrom

theSearch

oftheCam

bridge

Structural

Databasea

rawdatac

refineddatad

functio

nalgroups

presentb

no.o

fstructures

homosynthons

heterosynthons

no.o

fstructures

homosynthons

heterosynthons

range/Å

mean

distance/Å

OH

8009

1023

(13%

)3490

316(9%)

2.41−3.04

2.805(2)

OH

andN

arom

580

62(11%

)302(52%

)160

11(7%)

86(54%

)2.51−3.07

2.735(3)

OH

andCONH

2119

66(55%

)CONH

2···OCNH

238

(32%

)OH···OCNH

244

11(32%

)CONH

2···OCNH

226

(76%

)OH···OCNH

22.54−3.03

2.707(3)

7(6%)OH···OH

31(24%

)OH···NH

2CO

2(9%)OH···OH

24(70%

)OH···NH

2CO

2.62−3.09

2.812(4)

OH

andOC

2592

269(10%

)OH···OH

1136

(44%

)OH···OC

396

26(7%)OH···OH

166(42%

)OH···OC

2.42−3.04

2.745(2)

OH

andCOOH

683

230(33%

)COOH···OC(O

H)

387(56%

)OH···OC(O

H)

8635

(41%

)COOH···OC(O

H)

58(67%

)OH···OC(O

H)

2.38−3.02

2.672(3)

23(4%)O

H···OH

362(53%

)OH···OHCO

6(7%)OH···OH

55(64%

)OH···OHCO

2.61−3.08

2.812(2)

OH

andNH

2427

28(7%)OH···OH

179(42%

)OH···NH

2225

13(6%)OH···OH

101(64%

)OH···NH

22.56−3.07

2.848(4)

aThe

followingparameterswereused

assearch

criteria:3Dcoordinatesdeterm

ined,onlyorganiccrystalstructureswith

R≤7.5%

wereincluded

inthesearch

whilestructures

with

disorderanderrors,and

polymericspecieswereexcluded.T

hesearch

wasperformed

onCSD

Version

5.33

(Novem

ber2011)+2

updates.bAsearch

formolecules

containing

thespecified

functio

nalgroup(s)eitherexclusivelyor

inconjunctionwith

othergroups.cThe

rawsearch

results

referto

thespecified

functio

nalgroup(s)

includingoccurrenceswhere

othergroups

may

bepresent.dThe

refinedsearch

results

referto

the

specified

functio

nalgroup(s)

intheabsenceof

othergroups

intheimmediate

vicinity.

Crystal Growth & Design Article

dx.doi.org/10.1021/cg300689m | Cryst. Growth Des. XXXX, XXX, XXX−XXXE

materials are absent regardless of the solvent used, a differentsolid phase is produced with each solvent. However, it has notbeen possible to obtain suitable single crystals of these newphases to study the detailed hydrogen bonding arrangement(see Supporting Information for experimental PXRD data).The DSC data of the co-crystals reveal single sharp

endotherms at 157.9, 113.1, and 116.6 °C for products 1−3respectively. These endotherms, which correspond to themelting point of the solids, occur at significantly differenttemperatures to those of paracetamol (168−172 °C) or the co-crystal former (trans-1,4-diaminocyclohexane, 68−72 °C; trans-1,2-di-(4-pyridyl)-ethylene, 148−152 °C; 1,2-bis(4-pyridyl)-ethane, 110−112 °C), indicating the formation of co-crystalsand not simple physical mixtures. All three co-crystals describedhere have melting temperatures reduced from that ofparacetamol, suggesting that the cohesive energy of co-crystals1−3 is reduced from that of pure paracetamol form I.Full crystallographic details for crystal structure determi-

nations of the co-crystals 1−3 are presented in Table 3. The

local coordination and atom numbering scheme for each of thethree co-crystals is presented in Figure 3. Selected bonddistances and angles are presented as Supporting Informationwhile geometric parameters of the hydrogen bonds are reportedin Table 3, together with those for paracetamol. Co-crystal (1)contains one molecule of paracetamol and half a molecule oftrans-1,4-diaminocyclohexane in the asymmetric unit (Figure3a). N(2)−H(2)···O(2) hydrogen bonding interactionsbetween paracetamol molecules lead to the formation ofinfinite chains, which may be denoted C(4), directed along[100] (Figure 4)]. These chains are then cross-linked in the

[110] direction through O(1)−H(1)···N(1) and N(1)−H(16)···O(1) hydrogen bonding interactions to create a two-dimensional (2D) hydrogen bonded network (Figure 4). Alongwith several weak interactions there is an additional N(1)−H(15)···O(2) interaction between the 2D networks whichextend the structure into the third dimension.Co-crystal (2) contains in the asymmetric unit two molecules

of paracetamol, one molecule of trans-1,2-di-(4-pyridyl)-ethylene, and one molecule of methanol (Figure 3b). Pairs ofcrystallographically distinct paracetamol molecules are linked

Table 3. Crystallographic Data for Co-Crystals 1−3

1 2 3

chemicalformula

C11H16N2O2 C29H32N4O5 C40H42N6O4

formulaweight

208.26 516.59 670.8

crystal system triclinic triclinic triclinica (Å) 5.0174(5) 9.0546(2) 8.9707(4)b (Å) 8.8907(9) 10.6241(2) 12.8633(6)c (Å) 12.7562(1) 14.3162(3) 16.8305(2)α (°) 106.21(4) 82.97(3) 106.06(2)β (°) 91.40(5) 89.69(3) 95.27(2)γ (°) 90.15(5) 73.85(3) 95.13(3)space group P1 P1 P1V/Å3 546.2 (1) 1312.3 (5) 1844.8 (1)Z 2 2 2Nreflection/Nparameter

4141/136 4904/343 4389/469

ρcalc/g cm−3 1.266 1.307 1.208radiation type Mo Kα (λ =

0.710173 Å)Mo Kα (λ =0.710173 Å)

Mo Kα (λ =0.710173 Å)

T/K 100 100 293θ range/° 2−40 1−27 2−27range of h −8 to 8 −9 to 11 −11 to 11range of k −15 to 15 −17 to 18 −16 to 16range of l −18 to 21 −13 to 11 −16 to 21Rmerge 0.019 0.023 0.028R1 (%) 3.75 4.18 5.44WR2 (%) 4.24 4.34 5.67goodness-of-fit

1.072 1.117 1.102

Figure 3. Local coordination of the non-hydrogen atoms in the co-crystals of paracetamol with (a) trans-1,4-diaminocyclohexane, (1);(b) trans-1,4-di(4-pyridyl)ethylene, (2); 1,2-bis(4-pyridyl)ethane, (3).The displacement ellipsoids are drawn at the 50% probability level andatoms in the asymmetric unit are labeled.

Crystal Growth & Design Article

dx.doi.org/10.1021/cg300689m | Cryst. Growth Des. XXXX, XXX, XXX−XXXF

through N(4)−H(4)···O(2) intermolecular hydrogen bondinginteractions to form dimeric units, neighboring pairs of whichare connected through N(3)−H(2)···O(5) and O(5)−H(5)···O(4) hydrogen bonds involving the methanol solventincorporated in the co-crystal. This produces paracetamol-methanol chains directed approximately along [1 10]. The co-crystal former trans-1,2-di-(4-pyridyl)-ethylene serves to cross-link these chains through O(1)−H(1)···N(1) and O(3)−H(3)···N(2) hydrogen bonds (Figure 5, Table 4). Theneighboring aromatic rings of trans-1,2-di-(4-pyridyl)-ethyleneare arranged in a face to face35 stacking fashion with a slip angleof 18.2° with a centroid to centroid distance of around 3.32 Å.Aromatic interactions of this kind are known to generatesupramolecular synthons and have been exploited for co-crystalsynthesis.36 These paired units are then linked through severalweak interactions between the 2D networks to stabilize the co-crystal to extend into the third dimension.The asymmetric unit of the co-crystal (3) consists of four

independent molecules, two each of paracetamol and the co-crystal former 1,2-bis(4-pyridyl)ethane (Figure 3c). Analogousto co-crystals (1) and (2) the phenolic −OH and amidic −NH

of paracetamol are involved in hydrogen bonding. While in co-crystals (1) and (2) it is only the phenolic −OH that ishydrogen bonded to the CCF, here the amidic N−H is alsohydrogen bonded to the CCF. However, the carbonyl groups ofparacetamol are not involved in conventional hydrogen bondsbut participate in weak C−H···O interactions betweenmolecules related by an inversion center. The paracetamolmolecules are linked to two 1,2-bis(4-pyridyl)ethane molecules,through O(1)−H(1)···N(4) and N(6)−H(6)···N(3) interac-

Figure 4. The structure of cocrystal (1) viewed along [001]. Key: asfor Figure 1.

Figure 5. The structure of cocrystal (2) viewed along [001]. Key: as for Figure 1.

Table 4. Geometric Parameters for the Hydrogen Bonds inParacetamol (PAC) and the Co-Crystals 1−3

compound hydrogen bond

d(H···A)(Å)

d(D···A)(Å)

θ (D−H···A)(°)

(H···A) vdWcutoff

contraction(%)

PAC N−H···O 2.00 2.91 163.8 26PAC O−H···OC 1.77 2.66 166.1 36(1) N(2)−

H(2)···O(2)2.06 2.96 174.1 23

(1) O(1)−H(1)···N(1)

1.81 2.68 169.9 33

(1) N(1)−H(16)···O(1)

2.26 3.15 169.9 17

(1) N(1)−H(15)···O(2)

2.32 3.16 158.0 15

(2) O(1)−H(1)···N(1)

1.85 2.78 177.9 33

(2) N(3)−H(2)···O(5)

1.93 2.83 162.9 29

(2) O(3)−H(3)···N(2)

1.85 2.71 169.8 33

(2) N(4)−H(4)···O(2)

1.99 2.90 173.8 27

(2) O(5)−H(5)···O(4)

1.76 2.72 173.8 35

(3) O(1)−H(1)···N(4)

1.75 2.71 166.6 36

(3) N(6)−H(6)···N(3)

2.02 2.99 166.4 26

(3) O(3)−H(44)···N(2)

1.77 2.75 175.8 35

(3) N(5)−H(45)···N(1)

1.98 2.96 175.5 28

aD refers to the proton donor, A refers to the proton acceptor, d aredistances in Å. The atom numbering scheme is that presented inFigure 3.

Crystal Growth & Design Article

dx.doi.org/10.1021/cg300689m | Cryst. Growth Des. XXXX, XXX, XXX−XXXG

tions in one crystallographically independent molecule andO(3)−H(44)···N(2) and N(5)−H(45)···N(1) interactions inthe other (Table 3). This produces two crystallographically-distinct stepped chains of molecules (Figure 6). The CH2groups in one of the crystallographically independent 1,2-bis(4-pyridyl)ethane molecules were found to be disordered over twosites and were refined anisotropically with occupancies of 0.47and 0.53. Several weak C−H···O interactions between theseone-dimensional chains extend the structure into a three-dimensional structure.The paracetamol molecule contains two hydrogen bond

donors; the phenolic O−H and the amidic N−H groups andtwo acceptors; the amidic and phenolic oxygen atoms. Areasonable expectation would be for the strongest hydrogenbonds to be formed between the best donor and best acceptor.During co-crystal formation the donors and acceptors ofparacetamol are in competition with those of the co-crystalformer and the structure adopted is that which optimizes thedonor−acceptor interactions in the co-crystal. Geometricparameters of the hydrogen bonds in co-crystals 1−3 arepresented in Table 3. The strong O−H···OC interactions(d(D···A) = 2.66 Å) present in paracetamol are absent in theco-crystals. However, they are replaced by O−H···Ninteractions which operate over a similar distance and aretherefore of comparable strength. This is supported by thesimilarity in the contraction of the H···A van der Waals’ cutoffand red shift of the ν(D-H) stretching frequencies (vide infra).The geometric and spectroscopic data are consistent with thestrongest hydrogen bonds in the co-crystals resulting fromthese O−H···N interactions. Consistent with the suggestions ofEtter37 all of the remaining proton donors and acceptors alsoparticipate in hydrogen bond formation, as evidenced in allthree co-crystals by the N−H···OC interactions betweenparacetamol molecules, which are of similar strength to those inparacetamol itself. Co-crystal (2) has additional hydrogenbonds N−H···O and O−H···OC due to the incorporation ofsolvent molecules. This suggests a hierarchy of hydrogen bonds

in the co-crystals 1−3 which is in accord with the empiricalrules proposed by Etter.37

Complexes with hydrogen bonds are stabilized by electro-static, polarization, induction, and dispersion energy terms.From an electrostatic viewpoint, the interaction may beconsidered to have contributions from interactions betweenmonopoles, dipoles, quadrupoles, etc., each of which has adifferent characteristic cutoff in terms of distance. Onecommon means of assigning the character of the hydrogenbond is the contraction, with respect to the sum of the van derWaals’ radii, of the H···A distance. Proton-centered hydrogenbonds typically exhibit a contraction of between 40 and 55% ofthe sum of the van der Waals’ radii.38 In co-crystals 1−3 the vander Waals’ cutoff contraction lies between 20 and 40% (Table3), which suggests these hydrogen bonds may be classed asbeing of moderate strength38 in which the electrostaticinteraction terms are dominant. It has been empiricallyestablished that hydrogen bonds become increasingly linearwith increasing strength, as linearity optimizes the electrostaticcomponent (dipole−monopole, dipole−dipole) of the inter-action. The electrostatic contribution to the intermolecularinteraction tends to dominate and imparts directionality onhydrogen bonds. This is particularly significant in short andproton-centered hydrogen-bonds, whose typical angular rangeis 160 ≤ θ/° ≤ 180, but also applies to longer bonds ofelectrostatic nature. According to Reed et al.,39 charge transferbetween donor and acceptor groups has a strong influence onthe geometrical and spectral properties of the co-crystals. Inparticular, in a D−H···A hydrogen bond there is charge transferfrom lone pairs or π-molecular orbitals of the electron donor(proton acceptor) to the antibonding orbitals of the D−Hbond of the electron acceptor (proton donor). An increase ofelectron density in the antibonding orbitals elongates the D−Hbond, which causes the red-shift of the D-H stretchingfrequency. Therefore charge transfer and hence hydrogen-bond formation may be inferred from comparison ofcharacteristic vibrational frequencies of a free R-D-H group

Figure 6. The structure of co-crystal (3). Key: as Figure 1.

Table 5. Bond Stretching Frequencies in Paracetamol and the Three Co-Crystals, Determined by Infrared Spectroscopy

assignment paracetamol in gas-phase42 paracetamol Form I (1) (2) (3)

ν(O−H)/cm−1 3653 3160 3326 3070 3195ν(N−H)/cm−1 3465 3324 3326, 3273 3290 3246ν(C−H)aromatic/cm−1 3034 2895 3049 2870 2950ν(CO)/cm−1 1721 1651 1636 1653 1673

Crystal Growth & Design Article

dx.doi.org/10.1021/cg300689m | Cryst. Growth Des. XXXX, XXX, XXX−XXXH

with that of the same group in the presence of a hydrogen bondacceptor. The characteristic vibrational frequencies of the threeco-crystals 1−3 and of crystalline paracetamol form I may becompared with those observed in the gas phase infraredspectrum of paracetamol (Table 5). The frequencies of theν(O−H), ν(N−H), and ν(CO) stretching vibrations in co-crystals 1−3 and in paracetamol form I are red-shifted by up to16% relative to the equivalent vibrations in the gas-phasespectrum of paracetamol. From the crystal structure determi-nation, it can be deduced that the lone pair of the nitrogenatom of each of the co-crystal formers is aligned along the axisof the O−H bond of the hydroxyl group of paracetamol,thereby facilitating charge transfer between the donor andacceptor pair. Therefore, we conclude that both electrostaticand charge transfer terms influence the directional behavior ofthe hydrogen bonds in co-crystals 1−3.In all three co-crystals presented here, the structural and

spectroscopic data indicate that the hydrogen bonds formedbetween the API and the CCF are of comparable strength tothose originally present in the API. All three co-crystals studiedhere have the same, i.e., O−H···N supramolecular hetero-synthon between the API and CCF. However in structure (3)in addition to the O−H···N there is a N−H···N supramolecularinteraction observed between the API and CCF. Ourinvestigations show that difunctional CCFs, able to formchain-like structural motifs through distinct hydrogen bonds,are necessary for co-crystal formation with paracetamol. This isin accordance with previously reported CCFs in multi-component crystalline forms of paracetamol. Although CSDanalysis suggests carboxylic acids and amides as potential co-crystal formers, our experiments demonstrate that they are noteffective for paracetamol. This may be due to the self-complementary hydrogen bonding nature of these functionalgroups, which favors homosynthon formation. A hierarchicalarrangement between the best donor of paracetamol to the bestacceptor in CCF is observed in all the co-crystals presentedhere and is indeed observed in almost all of the previouslyreported co-crystal forms of paracetamol. Of the reported co-crystals of paracetamol, only those involving theophylline andbipyridine29 do not follow the hierarchical arrangement ofhydrogen bonding. In the case of the paracetamol-bipyridineco-crystal, strong π−π interactions (at a distance of ca. 3.4 Å)between the pyridine moieties of bipyridine stabilize the crystalstructure in co-operation with the hydrogen bonds. In the caseof the paracetamol-theophylline co-crystal, the best acceptor oftheophylline, the imidazole nitrogen atom, does not participatein conventional hydrogen bonds. The two molecules arearranged such that they form flat hydrogen bonded sheets.A generally accepted guideline40 is that proton transfer will

occur from an acid to a base when the difference in pKa (whereΔpKa refers to the difference between the pKa value of aprotonated base and that of the acid) is >3.5, and a neutral co-crystal will form when the ΔpKa < 0. A co-crystal to saltcontinuum30d exists between 0 ≤ ΔpKa ≤ 3.5. The ambiguity inthis range of ΔpKa values may be a problem when applyingcrystal engineering principles to co-crystal formation, as thecharge separation resulting from salt formation leads to thedominance of electrostatic interactions over the noncovalentinteractions, on which crystal engineering is based. Moreover,the prediction of the degree of proton transfer in the solid-stateform based on the pKa difference remains imprecise and mayalso show variations with the solvent of crystallization.Calculated pKa values

41 of the CCFs used in this work indicate

that neither the amine nor pyridyl CCFs investigated aresufficiently basic to form a salt with paracetamol throughcomplete proton transfer. Calculated values ΔpKa of <0 arethus consistent with the formation of a neutral complex.

■ CONCLUSIONSAnalysis of structural data in the Cambridge StructuralDatabase resulted in the identification of a range of potentialco-crystal formers compatible with paracetamol. Subsequentscreening using solvent-drop grinding in conjunction with high-throughput PXRD led to the discovery of three new co-crystalsof paracetamol. All the three co-crystals presented here utilizethe O−H···N supramolecular heterosynthon. The ability ofphenols to form this heterosynthon is consistent with theresults of the analysis of the structural database, suggesting theapproach may be applicable to a wider range of phenolderivatives. Comparison of the geometric and spectroscopicparameters of the hydrogen-bonds in co-crystals 1−3 withthose for paracetamol reveals that they are of similar strength,irrespective of the chemical identity of the synthon. The studyshows that a balance between the retrosynthetic approach(synthon method) and database screening of supramolecularsynthons provides a useful approach to the targeted synthesis.

■ ASSOCIATED CONTENT*S Supporting InformationCrystallographic information files, CSD search details, DSC,NMR and IR spectra, powder diffraction patterns of screeningexperiments using trans-1,2-di-(4-pyridyl)-ethylene as CCF,comparison plot between single crystal simulated and bulkpowder patterns, calculated pKa values of paracetamol andCCFs. This information is available free of charge via theInternet at http://pubs.acs.org/.

■ AUTHOR INFORMATIONCorresponding Author*Tel: +44 131 451 8034. Fax: +44 (0)131 451 3180. E-mail: a.v.powell@hw.ac.uk.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThe authors wish to thank the Scottish Funding Council-SPIRIT programme and Heriot-Watt University for funding.

■ REFERENCES(1) Bond, A. D. CrystEngComm 2007, 9, 833.(2) Buck, J. S.; Ide, W. S. J. Am. Chem. Soc. 1931, 53, 2784.(3) Anderson, J. S. Nature 1937, 140, 583.(4) Hall, B.; Devlin, J. P. J. Phys. Chem. 1967, 71, 465.(5) Desiraju, G. R. CrystEngComm 2003, 5, 466.(6) Jack, J. D. CrystEngComm 2003, 5, 506.(7) Byrn, S. R.; Pfeiffer, R. R.; Stowell, J. G. Solid State Chemistry ofDrugs, 2nd ed.; SSCI Inc.: IN, 1999.(8) (a) Bernstein, J. Polymorphism in Molecular Crystals; ClarendonPress: Oxford, 2002. (b) Davey, R. J. Chem. Commun. 2003, 13, 1463.(c) Rogers, R. D. Cryst. Growth Des. 2004, 4, 1085.(9) (a) Vishweshwar, P.; Beauchamp, D. A.; Zaworotko, M. J. Cryst.Growth Des. 2006, 6, 2429. (b) Mondal, R.; Howard, J. A. K.; Banerjee,R.; Desiraju, G. R. Cryst. Growth Des. 2006, 6, 2507. (c) Banerjee, R.;Bhatt, P. M.; Desiraju, G. R. Cryst. Growth Des. 2006, 6, 1468.(d) Barnett, S. A.; Tocher, D. A.; Vickers, M. CrystEngComm 2006, 8,313. (e) Hosokawa, T.; Datta, S.; Sheth, A. R.; Brooks, N. R.; Young,V. G.; Grant, D. J. W. Cryst. Growth Des. 2004, 4, 1195.

Crystal Growth & Design Article

dx.doi.org/10.1021/cg300689m | Cryst. Growth Des. XXXX, XXX, XXX−XXXI

(10) (a) Vishweshwar, P.; McMahon, J. A.; Bis, J. A.; Zaworotko, M.J. J. Pharm. Sci. 2006, 95, 499. (b) Vishweshwar, P.; McMahon, J. A.;Peterson, M. L.; Hickey, M. B.; Shattock, T. R.; Zaworotko, M. J.Chem. Commun. 2005, 4601. (c) Fleischman, S. G.; Kuduva, S. S.;McMahon, J. A.; Moulton, B.; Bailey Walsh, R. D.; Rodriguez-Hornedo, N.; Zaworotko, M. J. Cryst. Growth Des. 2003, 3, 909.(d) Reddy, L. S.; Babu, N. J.; Nangia, A. Chem. Commun. 2006, 1369.(e) Childs, S. L.; Hardcastle, K. I. Cryst. Growth Des. 2007, 7, 1291.(f) Childs, S. L.; Chyall, L. J.; Dunlap, J. T.; Smolenskaya, V. N.;Stahly, B. C.; Stahly, G. P. J. Am. Chem. Soc. 2004, 126, 13335.(11) (a) Stahl, P. H.; Nakano, M. In Handbook of PharmaceuticalSalts: Properties, Selection, and Use; Stahl, P. H.; Wermuth, C. G., Eds.;Wiley-VCH: New York, 2002; Ch. 4. (b) Berge, S. M.; Bighley, L. D.;Monkhouse, D. C. J. Pharm. Sci. 1977, 66, 1.(12) Oswald, I. D. H.; Allan, D. R.; McGregor, P. A.; Motherwell, W.D. S.; Parsons, S.; Pulham, C. R. Acta Crystallogr., Sect. B 2002, 58,1057.(13) Almarsson, O; Zaworotko, M. J. Chem. Commun. 2004, 1889.(14) Rodriguez-Spong, B.; Price, C. P.; Jayasankar, A.; Matzger, A. J.;Rodriguez-Hornedo, N. Adv. Drug Delivery Rev. 2004, 56, 241.(15) Blagden, N.; De Matas, M.; Gavan, P. T.; York, P. Adv. Drug.Delivery Rev. 2007, 59, 617.(16) Desiraju, G. R. Crystal Engineering. The Design of Organic Solids;Elsevier: Amsterdam, 1989.(17) Pepinsky, R. Phys. Rev. 1955, 100, 971.(18) Schmidt, G. M. J. Pure Appl. Chem. 1971, 27, 647.(19) Thomas, J. M. Phil. Trans. R. Soc. Lond. A 1974, 277, 251.(20) Dunitz, J. D. Pure Appl. Chem. 1991, 63, 177.(21) Desiraju, G. R. Angew. Chem., Int. Ed. 1995, 34, 2311.(22) Shattock, T. S.; Arora, K. K.; Vishweshwar, P.; Zaworotko, M. J.Cryst. Growth Des. 2008, 8, 4533.(23) The method of search for existing cocrystals: The search wasconducted using the parameters option “number of chemical units” asmore than or equal to two. Hydrates and solvates were excluded fromthe search.(24) (a) Allen, F. H.; Raithby, P. R.; Shields, G. P.; Taylor, R. Chem.Commun. 1998, 1043. (b) Allen, F. H.; Motherwell, W. D. S.; Raithby,P. R.; Shields, G. P.; Taylor, R. New. J. Chem. 1999, 23, 25.(25) (a) Drebushchak, T. N.; Boldyreva, E. V. Z. Kristallogr. 2004,219, 506. (b) Wilson, C. C. Z. Kristallogr. 2000, 215, 693.(26) Burger, A. Acta Pharm. Technol 1982, 28, 1.(27) Perrin, M-A; Neumann, M. A.; Elmaleh, H.; Zaska, L. Chem.Commun. 2009, 3181.(28) Bernstein, J.; Davis, R. E.; Shimoni, L.; Chang, N. L. Angew.Chem. Int. Ed. 1995, 34, 1555.(29) Karki, S.; Friscic, T.; Fabian, L.; Laity, P. R.; Day, G. M; Jones,W. Adv. Mater. 2009, 21, 3905.(30) (a) Oswald, I. D. H.; Allan, D. R.; McGregor, P. A.; Motherwell,W. D. S; Parsons, S.; Pulham, C. R.. Acta Crystallogr. Sect. B 2002, 58,1057. (b) Andre, V.; Piedade, M. F. M; da; Duarte, M. T..CrystEngComm 2012, 14, 5005. (c) Vrcelj, R. M.; Clark, N. I. B.;Kennedy, A. R.; Sheen, D. B.; Shepherd, E. E. A.; Sherwood, J. N. J.Pharm. Sci. 2003, 92, 2069. (d) Oswald, I. D. H.; Motherwell, W. D. S;Parsons, S.; Pulham, C. R. Acta Crystallogr. Sect. E 2002, 58, O1290.(d) Oswald, I. D. H..; Pulham, C. R. CrystEngComm 2008, 10, 1114.(e) Childs, S. L.; Stahly, G. P.; Park, A. Mol. Pharmaceutics 2007, 4,323.(31) Trask, A. V.; Motherwell, W. D. S; Jones, W. Chem Commun.2004, 890.(32) APEX-2 software, Version 1.27; Bruker AXS Inc.: Madison,Wisconsin, USA, 2005.(33) Sheldrick, G. M. Acta Crystallogr., Sect. A 2008, 64, 112.(34) Betteridge, P. W.; Carruthers, J. R.; Cooper, R. I.; Prout, K.;Watkin, D. J. J. Appl. Crystallogr. 2003, 36, 1487.(35) Collings, J. C.; Roscoe, K. P.; Thomas, R. Ll.; Batsanov, A. S.;Stimson, L. M.; Howard, J. A. K.; Marder, T. B. New J. Chem. 2001, 25,1410.(36) Salonen, L. M.; Ellermann, M.; Diedrich, F. Angew. Chem., Int.Ed. 2011, 50, 4808.

(37) Etter, M. C. Acc. Chem. Res. 1990, 23, 120.(38) Gilli, G.; Gilli, P. The Nature of the Hydrogen Bond Outline of aComprehensive Hydrogen Bond Theory; Oxford University Press:Oxford, 2009.(39) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. Rev. 1988, 88,899.(40) Johnson, S. L.; Rumon, K. A. J. Phys. Chem. 1965, 69, 74.(41) The SPARC program (http://archemcalc.com/sparc) usescomputational algorithms based on fundamental chemical structuretheory to estimate a variety of reactivity parameters. For someexamples of the use of the SPARC program for pKa calculations, see(a) Erdemgil, F. Z.; Sanli, S.; Sanli, N.; Ozkan, G.; Barbosa, J.;Guiteras, J.; Beltran, J. L. Talanta 2007, 72, 489. (b) Hilal, S. H.;Karickhoff, S. W.; Carreira, L. A. Quant. Struct. Act. Rel. 1995, 14, 348.(c) Hilal, S. H.; ElShabrawy, Y.; Carreira, L. A.; Karickhoff, S. W.;Toubar, S. S. Talanta 1996, 43, 607. (d) Hilal, S. H.; Carreira, L. A.;Baughman, G. L.; Karickhoff, S. W.; Melton, C. M. J. Phys. Org. Chem.1994, 7, 122.(42) NIST Standard Reference Database 35, NIST/EPA Gas-PhaseInfrared Database; National Institute of Standards and Technology:Gaithersburg, MD.

Crystal Growth & Design Article

dx.doi.org/10.1021/cg300689m | Cryst. Growth Des. XXXX, XXX, XXX−XXXJ